De novo transcriptome sequencing of axolotl blastema for identification of differentially expressed genes during limb regeneration
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Wu et al. BMC Genomics 2013, 14:434
http://www.biomedcentral.com/1471-2164/14/434
RESEARCH ARTICLE Open Access
De novo transcriptome sequencing of axolotl
blastema for identification of differentially
expressed genes during limb regeneration
Cheng-Han Wu1†, Mong-Hsun Tsai1,5†, Chia-Chuan Ho2, Chien-Yu Chen3* and Hsuan-Shu Lee1,4,5,6*
Abstract
Background: Salamanders are unique among vertebrates in their ability to completely regenerate amputated limbs
through the mediation of blastema cells located at the stump ends. This regeneration is nerve-dependent because
blastema formation and regeneration does not occur after limb denervation. To obtain the genomic information of
blastema tissues, de novo transcriptomes from both blastema tissues and denervated stump ends of Ambystoma
mexicanum (axolotls) 14 days post-amputation were sequenced and compared using Solexa DNA sequencing.
Results: The sequencing done for this study produced 40,688,892 reads that were assembled into 307,345
transcribed sequences. The N50 of transcribed sequence length was 562 bases. A similarity search with known
proteins identified 39,200 different genes to be expressed during limb regeneration with a cut-off E-value
exceeding 10-5. We annotated assembled sequences by using gene descriptions, gene ontology, and clusters of
orthologous group terms. Targeted searches using these annotations showed that the majority of the genes were
in the categories of essential metabolic pathways, transcription factors and conserved signaling pathways, and
novel candidate genes for regenerative processes. We discovered and confirmed numerous sequences of the
candidate genes by using quantitative polymerase chain reaction and in situ hybridization.
Conclusion: The results of this study demonstrate that de novo transcriptome sequencing allows gene expression
analysis in a species lacking genome information and provides the most comprehensive mRNA sequence resources
for axolotls. The characterization of the axolotl transcriptome can help elucidate the molecular mechanisms
underlying blastema formation during limb regeneration.
Keywords: Ambystoma mexicanum, Axolotl, Transcriptome, Blastema, De novo transcriptome sequencing
Background amputation site within several hours [2-4]. After the
Ambystoma mexicanum (axolotl), one of over 500 spe- nerves and wound epidermis contact each other, the epi-
cies of salamander, can completely reconstitute lost dermis overlying the axon ends thickens, forming an ap-
limbs after amputation. The amputation of limbs results ical epithelial cap [5]. Fibroblasts from the surrounding
in the formation of blastemas in the stump ends. These tissue simultaneously migrate to the amputation site
blastemas contain undifferentiated cells capable of grow- under the apical epithelial cap. These fibroblasts prolifer-
ing and developing into new limbs exactly as they were ate to form a mass of undifferentiated cells that subse-
before amputation [1]. In the early phase of regener- quently develops into the new limb. In the absence of
ation, growing wound epithelium and epidermis cover functional nerves, an apical epithelial cap and blastema
the ends of the truncated nerves and the surface of the cannot be formed on the amputation stump [6]. Instead,
denervated limbs undergo a wound-healing response
* Correspondence: cychen@mars.csie.ntu.edu.tw; benlee@ntu.edu.tw
post-amputation, and do not regenerate [7,8].
†
Equal contributors In past several years, next-generation sequencing (NGS)
3
Department of Bio-Industrial Mechatronics Engineering, National Taiwan technology has become a cutting-edge approach for high-
University, Taipei 106, Taiwan
1
Institute of Biotechnology, National Taiwan University, Taipei 106, Taiwan
throughput sequence determination. This technology has
Full list of author information is available at the end of the article dramatically improved the efficiency and speed of gene
© 2013 Wu et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Wu et al. BMC Genomics 2013, 14:434 Page 2 of 12
http://www.biomedcentral.com/1471-2164/14/434
discovery in many studies [9,10], and has significantly sequences, including hundreds of developmental genes and
accelerated and improved the sensitivity of gene expres- wound-healing genes were identified. The gene expression
sion profiling. For example, studies in the field of human profiles of a regenerating blastema and a non-regenerating
[11] and Arabidopsis [12] transcriptomics have made re- denervated limb stump, 14 days post-amputation, were
markable progress. Studies using transcriptome sequen- compared using differential gene expression analysis. A
cing for organisms with complete genome sequencing list of genes significantly overexpressed in normal regen-
have confirmed that the short-read products of NGS can erating blastema was obtained from the results of the
be effectively assembled and used for gene discovery and analysis. A subset of genes from this list was verified
comparison of gene expression profiles. using quantitative polymerase chain reaction (qPCR) and
National Center for Biotechnology Information (NCBI) in situ hybridization.
[13-15] and Sal-Site [16] have already made available
many cDNA sequene databases for axolotl. However, Results and discussion
there still exist many unannotated protein-coding genes Illumina NGS and read assembly
or non-coding RNA sequences that have not been sam- To obtain the gene expression profiles of an axolotl blas-
pled previously and are not present in existing cDNA li- tema and a non-regenerating denervated limb stump,
braries. Recent reports have also suggested that more mRNA samples of both types of tissue, 14 days post-
Ambystoma to human non-redundant (nr) orthologous amputation, were prepared and sequenced using an
sequences remain to be discovered [15]. Moreover, se- Illumina sequencing platform. We performed a single se-
quence coverage of transcripts is highly variable between quencing run for each of the 2 tissues. After cleaning and
different cDNA libraries. With more available cDNA se- quality assurance, we obtained 40 million 100-bp reads
quences, the overall sequence coverage of axolotl tran- from both samples (Table 1). These raw reads were ran-
scripts will be improved. Although previous studies have domly clipped into 21-mers for sequence assembly using
highlighted the usefulness of cDNA sequencing for the SOAP de-novo [22] to yield 2,920,951 contigs (Table 1)
discovery of candidate genes in the absence of a genome with a mean contig size of 117 bp. The size distribu-
sequence database, a comprehensive description of the tion of these contigs is shown in Additional file 1. In
full spectrum of genes expressed in axolotl blastemas is a previous report [14], Monaghan et al. generated over
still lacking. To our knowledge, the genome sequencing 1.7 million reads and approximately 400,000 unique se-
of any salamander species has not been completed. quences with a mean contig size of 215 bp for axolotl
Several studies have used highly parallel 454 pyro- from a broader range of regeneration stages using 454
sequencing to identify axolotl sequences which are used pyrosequencing. Using paired-end joining and gap-filling,
to generate a large-scale feature axolotl microarray we further assembled these contigs into 307,345 tran-
[14,15,17,18]. However, 454 pyrosequencing has relatively scribed sequences with a mean size of 373 bp including
lower overall transcriptome coverage when compared to 20,504 transcribed sequences larger than 1000 bp (Table 1).
Illumina/Solexa platforms [19-21]. Several recent studies After clustering using TIGR Gene Indices clustering tools
have employed the Illumina/Solexa platform to offer a far (TGICL) [23], the 307,345 transcribed sequences generated
greater coverage than 454 pyrosequencing [19-21]. How- 116,787 unigenes with a mean size of 529 bp (Table 1). In
ever, in the early stages of this platform, the majority of this report, the term “unigene” indicates a transcribed
Illumina sequence reads could not be matched to known
genes because of their short length. In general, 454
pyrosequencing had longer sequence reads whereas Table 1 Summary for Ambystoma mexicanum
Illumina sequencing had shorter, but more numerous transcriptome
paired ends read [19-21]. Currently, the latest develop- Total number of reads 40,688,892
ments in 454 and Illumina technologies offer higher reso- Total base pairs (bp) 4,862,000,340
lution and are relatively consistent with each other. With Average read length 100 bp
improved quality and longer reads, the higher coverage Total number of contigs 2,920,951
from Illumina technologies allows for the identification
Mean length of contigs 117 bp
of low-abundance genes not detected in earlier studies of
limb regeneration based on 454 pyrosequencing. There- Total number of transcribed sequences 307,345
fore, Illumina platforms are well suited for gene discovery Mean length of transcribed sequences 373 bp
and promising insights into axolotl limb regeneration. Transcribed sequence N50 562 bp
The de novo transcriptome sequencing of axolotl Total unigenes 116,787
blastema in this study produced over 4 billion bases of Mean length of unigenes 529
high-quality cDNA sequences, which were assembled and
Unigenes with E-value < 10-5 39,200
annotated without a reference genome. 116,787 distinctWu et al. BMC Genomics 2013, 14:434 Page 3 of 12 http://www.biomedcentral.com/1471-2164/14/434 sequence that matches no other transcribed sequence and The sequence similarity distribution has a comparable has distinct sequences. pattern, with 24.8% of the sequences having a similarity The data sets are available at the NCBI Short Read higher than 80%, and 75.2% of the sequences having a Archive (SRA) with the accession number: SRA064951. similarity ranging from 14% to 80% (Figure 1C). Annotation of predicted proteins Gene ontology and classification of clusters of For annotation, BLASTx was used to compare unigenes orthologous groups against the NCBI nr database by using a cut-off Gene ontology (GO) assignments were used to classify E-value of 10-5. Using this approach, 39,200 genes (33.6% the functions of the predicted axolotl genes. Based on of unigenes) were found to be above the E-value cut-off sequence homology, 15,633 sequences can be catego- (Additional file 2). Because of the relatively short length rized into 48 functional groups (Figure 2). In each of of unigene sequences (mean size of 529 bp) and lack of the 3 main categories of GO classification (i.e., biological genome reference in axolotl, most of the 77,587 (66.4%) process, cellular components, and molecular function), assembled sequences could not be matched to any “cellular process”, “cell part”, and “binding” terms are known genes. Figure 1A shows that the proportion of dominant, respectively. However, we found a few genes assembled sequences with matches in the nr database is in the categories: “virion”, “virion part”, “metallochaperone higher among the longer sequences. Specifically, se- activity”, and “electron carrier activity”. We also noticed a quences longer than 2000 bp had a match efficiency of high percentage of genes in the “developmental process” 80.6%, whereas the match efficiency decreased to ap- and “response to stimulus” categories, but only a few genes proximately 43.8% for sequences ranging from 500 to for “cell killing” and “antioxidant activity”. To further 1000 bp in length, and to 23.9% for sequences between evaluate the completeness of this transcriptome library and 200 and 500 bp in length (Figure 1A). The E-value distri- the effectiveness of the proposed annotation process, bution of the top matches in the nr database showed that we searched the annotated sequences for the genes 40% of the mapped sequences have strong homology involved in clusters of orthologous groups (COG) clas- (smaller than 1.0E-50), whereas 60% of the homologous sifications. Of the 39,200 matches to the nr database, sequences ranged from 1.0E-5 to 1.0E-50 (Figure 1B). 9744 sequences have a COG classification (Figure 3). Figure 1 Characteristics of homology search of Illumina sequences against the nr database. (A) Effect of query sequence length on the percentage of sequences for which significant matches were found. The proportion of sequences with matches (with a cut-off E-value of 1.0E-5) in the NCBI nr databases is greater among the longer assembled sequences. (B) E-value distribution of BLAST hits against the nr database for each unique sequence with a cut-off E-value of 1.0E-5. (C) Similarity distribution of the top BLAST hits for each sequence.
Wu et al. BMC Genomics 2013, 14:434 Page 4 of 12 http://www.biomedcentral.com/1471-2164/14/434 Figure 2 Histogram presentation of Gene Ontology classification. Results are summarized in three main categories: biological process, cellular component, and molecular function. The right y-axis indicates the number of genes in a category. The left y-axis indicates the percentage of a specific category of genes in the main category. Among the 25 COG categories, the cluster for “general “extracellular structures” (9, 0.049%), “nuclear structure” function prediction” represents the largest group (3785, (7, 0.049%), and “RNA processing and modification” (79, 20.7%), followed by “replication, recombination, and re- 0.43%) (Figure 3). To identify the biological pathways ac- pair” (1901, 10.4%) and “transcription” (1436, 7.9%). tive in the axolotl, we mapped unigenes to canonical ref- The following categories represent the smallest groups: erence pathways using the Kyoto Encyclopedia of Genes Figure 3 Histogram presentation of clusters of orthologous groups (COG) classification. Out of 39,228 nr hits, 9744 sequences have a COG classification among the 25 categories.
Wu et al. BMC Genomics 2013, 14:434 Page 5 of 12
http://www.biomedcentral.com/1471-2164/14/434
and Genomes (KEGG) [24]. In total, 24,095 sequences accuracy: Trinity [40], trans-ABySS [41], and SOAP de-
were assigned to 217 KEGG pathways. The pathways novo [42]. BLASTx was utilized to compare the assembled
with the most representation by the unique sequences contigs from the 3 respective assemblers with the protein
were starch and sucrose metabolism (2801 members, database of Xenopus laevis. The contigs with the best
11.62%), pathways in cancer (1126 members, 4.67%), and e-values were assigned to represent the potential axolotl
regulation of actin cytoskeleton (1029 members, 4.27%). genes. The fold change of the reads per kilo base of isotig
These annotations provide a valuable resource for inves- length per million mapped reads (RPKM) values between
tigating specific processes, functions, and pathways in the libraries from the two tissue types was calculated to
axolotl limb regeneration research. identify differentially expressed genes. The numbers of dif-
ferentially expressed genes for each assembler are: Trinity
Detection of growth factor and transcription factor 2651, trans-ABySS 2693, and SOAP de-novo 2833. Because
sequences related to limb regeneration in axolotls various assemblers deliver different sets of differentially
Fibroblast growth factors (FGFs) participate in salamander expressed genes, only genes homologous to a Xenopus
limb regeneration [25-29]. For example, fgf-8 is expressed laevis protein with corresponding contigs from all the 3 as-
in the basal layer of the apical epithelial cap [30]. Develop- semblers were selected to obtain more accurate results.
mental genes are also re-expressed in blastema mesenchy- With this strategy, we detected 1953 genes with differential
mal cells [31-36]. For example, homeobox A13, an autopod expression levels between blastema and denervated limb
marker in many vertebrates, appears in the distal region of stump libraries. Of these differentially expressed genes, 885
the blastema [32,37,38]. Therefore, sequences related to were up-regulated and 1068 were down-regulated in blas-
growth factors and transcription factors involved in regen- tema (Additional file 3). Among the up-regulated genes,
eration were analyzed and compared to sequences from anterior gradient rescues a denervated limb stump to be-
NCBI nucleotides and the EST database. As Table 2 shows, come regenerating again [43]; matrix metalloproteinase 13
this process identified a number of sequences that are (mmp13) is associated with metamorphic developmental
homologous to growth factors related to limb regeneration, processes in axolotl epidermis [44]; sox9 is essential for
such as transforming growth factor (tgf)-βs and fgfs, as well sclerotome development and cartilage formation [45]; and
as transcription factors involved in limb regeneration, such patched1 mediates hedgehog signaling into blastema [46].
as Homeobox (hox) genes. In total, this study identified 6 The set of differentially expressed genes from this study
tgf-β sequences. After removing redundant sequences, 3 was compared with previous sequencing projects using
different tgf-β isoforms, 17 different hox genes, and 7 dif- 454 pyrosequencing and microarray [14] (Additional
ferent fgfs were identified (Table 3). Four tgf-βs (tgf-β1, 2, 3, file 4). Only the data sets in the previous studies on tissues
and 5) have been identified previously in axolotl [39]. This 14 days after amputation, similar to this study, were chosen
study identifies 3 (tgf-β1, 2, 3) and represents a nearly for comparison. The unigenes that could not be annotated
complete collection of such genes in axolotl. These find- were excluded here. The Venn diagrams in Additional
ings confirm the high quality and high yield of the sequen- file 4 reveal overlaps between these 3 data sets. Among
cing data in this study, and may facilitate axolotl limb the overlapped genes, genes known to be necessary for
regeneration research. limb development and limb regeneration were identified
including wnt5a which is involved in Wnt/planar cell
Comparing gene expression profiles between blastema polarity signaling, and msx2 which plays important roles
and denervated limb stump in the development of limb buds.
To identify specific genes participating in limb regener-
ation, the differentially expressed genes between blastema Functional annotation of differentially expressed genes
and non-regenerating denervated limb stump were ana- To understand the functions of differentially expressed
lyzed using 3 different de novo assemblers to increase genes, the set of 1953 differentially expressed genes
Table 2 Comparison of the sequence numbers of transforming growth factor-βs, fibroblast growth factors, and hox
genes identified in this study and in an EST database
a b c
Gene name Number of sequences had Number of known sequences Number of known sequences in
a hit with nr database from NCBI nucleotide database an EST sequencing project [13]
Transforming growth factor-βs 6 1 1
Fibroblast growth factors 8 5 2
Hox genes 28 7 1
a Number of sequences obtained in this study that have a hit with the corresponding proteins in the NCBI nr database.
b Number of sequences from NCBI nr database hit in a that show homology with corresponding proteins from NCBI nucleotide database (to July 2012).
c Number of sequences that can be identified in an EST database established by an ambystoma mexicanum EST sequencing project [13].http://www.biomedcentral.com/1471-2164/14/434
Wu et al. BMC Genomics 2013, 14:434
Table 3 Identified transforming growth factor-βs, fibroblast growth factors, and Hox genes
Gene name Gene ID Length Subject ID Species E value
Transforming growth factor-beta-1 Unigene8146_All 3212 gi|160,334,206|gb|ABX24523.1| Ambystoma mexicanum 0
Transforming growth factor-beta-2 Unigene17537_All 2048 gi|155,966,093|gb|ABU41003.1| Anser anser 0
Transforming growth factor-beta-3 Unigene9879_All 866 gi|257,173|gb|AAB23575.1| Gallus gallus 5.00E-19
Fibroblast growth factor-2 Unigene114478_All 512 gi|15,216,301|dbj|BAB63249.1 Cynops pyrrhogaster 9.00E-80
Fibroblast growth factor-8 Unigene116687_All 1666 gi|12,024,873|gb|AAG45674.1|AF190448_1| Ambystoma mexicanum 1.00E-117
Fibroblast growth factor-9 Unigene82354_All 1419 gi|61,971,458|gb|AAX58114.1| Didelphis albiventris 1.00E-113
Fibroblast growth factor-10 Unigene32567_All 2235 gi|19,031,193|gb|AAK59700.1| Ambystoma mexicanum 1.00E-113
Fibroblast growth factor-12 Unigene43917_All 396 gi|238,815,007|gb|ACR 56700.1| Ovis aries 5.00E-44
Fibroblast growth factor-16 Unigene113330_All 434 gi|224,098,374|ref|XP_002195984.1| Taeniopygia guttata 7.00E-36
Fibroblast growth factor-17-like Unigene88737_All 636 gi|301,757,980|ref|XP_002914827.1| Ailuropoda melanoleuca 4.00E-43
HoxA3 Unigene18117_All 1271 gi|220,898,179|gb|ACL 81435.1| Latimeria menadoensis 1.00E-115
HoxB3 Unigene92817_All 203 gi|37,528,843|gb|AAQ92347.1| Pleurodeles waltl 4.00E-16
HoxD3 Unigene38161_All 710 gi|126,341,823|ref|XP_001362908.1| Monodelphis domestica 7.00E-72
HoxD4 Unigene74557_All 206 gi|301,128,902|emb|CBL59364.1| Scyliorhinus canicula 3.00E-06
HoxA7 Unigene15254_All 478 gi|6,754,234|ref|NP_034585.1| Mus musculus 1.00E-27
HoxB8 Unigene107183_All 283 gi|256,014,548|gb|ACU56828.1| Ichthyophis cf. kohtaoensis JMW-2009 2.00E-18
HoxD8 Unigene1050_All 1896 gi|7,271,820|gb|AAF44632.1|AF224263_2| Heterodontus francisci 6.00E-95
HoxA9 Unigene115485_All 641 gi|149,633,999|ref|XP_001509421.1| Ornithorhynchus anatinus 7.00E-72
HoxC9 Unigene108008_All 294 gi|220,898,209|gb|ACL81463.1| Latimeria menadoensis 4.00E-34
HoxD9 Unigene101797_All 242 gi|118,093,579|ref|XP_001234507.1| Gallus gallus 3.00E-25
HoxA10 Unigene8140_All 2019 gi|301,607,691|ref|XP_002933440.1| Xenopus (Silurana) tropicalis 1.00E-127
HoxC10 Unigene115613_All 670 gi|11,037,556|gb|AAG27630.1|AF298185_1| Ambystoma mexicanum 1.00E-101
HoxD10 Unigene114687_All 531 gi|194,043,944|ref|XP_001925011.1| Sus scrofa 2.00E-75
HoxC11a Unigene109718_All 323 gi|301,614,392|ref|XP_002936692.1| Xenopus (Silurana) tropicalis 3.00E-39
HoxC12 Unigene86529_All 681 gi|24,637,176|gb|AAN63593.1|AF434200_1| Pleurodeles walti 5.00E-68
HoxA13 Unigene115397_All 626 gi|220,898,176|gb|ACL81432.1| Latimeria menadoensis 1.00E-103
HoxD13 Unigene116723_All 1926 gi|301,612,439|ref|XP_002935723.1| Xenopus (Silurana) tropicalis 1.00E-123
Page 6 of 12Wu et al. BMC Genomics 2013, 14:434 Page 7 of 12
http://www.biomedcentral.com/1471-2164/14/434
which had corresponding contigs from all 3 assemblers up-regulated in blastema, qPCR results confirmed that 33
were mapped to terms in the GO biological process genes had higher expression levels in blastema (Figure 5A).
database and compared with the whole transcriptome Results also confirmed a >10-fold higher expression
background to identify genes involved in important bio- level for msx1 in blastema. Msx1 is involved in the
logical processes (Figure 4). Among all genes with GO regulation of dedifferentiation for mature myofibers.
assignment annotations, 1848 differentially expressed The morpholino-based suppression of Msx1 protein
genes between blastema and denervated limb stump librar- expression in myonuclei significantly inhibits the frag-
ies were annotated. Among the differentially expressed mentation and dedifferentiation of salamander myofibers
genes, genes involved in nitrogen and DNA metabolism [47]. In the initially developed limb bud in mice, msx1
were significantly up-regulated in blastema, such as glucose appears in the entire mesenchyme. However, this expres-
dehydrogenase, GPI mannosyltransferase, and DNA repli- sion is restricted to the most distal part of the limb bud
cation licensing factor mcm. These findings suggest that as development proceeds [48]. When msx1-negative
the cell proliferation rate in axolotl blastema is higher than proximal limb tissues are grafted into the distal portion
in denervated limb stumps, a finding consistent with that of the limb bud, msx1 is re-expressed in the mesenchyme
of a previous report [46]. of the transplanted tissues, which subsequently become
dedifferentiated [49]. The homeo box gene hoxd13 was
Validation of differentially expressed genes by qPCR shown to be expressed at higher levels in blastema. In
Among the genes classified as differentially expressed in mice, the hoxd13 gene plays a key role in axial skeleton
blastema, 47 regeneration-related genes were selected development and forelimb morphogenesis [50]. The
based on our interests and their differential expression present sequence analysis shows that fgfs were up-
levels were verified by qPCR. Amplification of an endogen- regulated in blastema, and this pattern was partly con-
ous gene, S21 ribosomal RNA, was used for normalization. firmed by showing up-regulation of fgf-8b and fgf-9 in
Among the 39 genes determined by sequence analysis as qPCR analysis. The expression patterns of fgfs in this
Figure 4 GO assignment of genes up- or down-regulated in balstema compared with the control non-regenerating limb stump. The
results of GO biological function assignment annotation of differentially expressed genes are summarized in two main categories: biological
process and molecular function.Wu et al. BMC Genomics 2013, 14:434 Page 8 of 12
http://www.biomedcentral.com/1471-2164/14/434
Figure 5 qPCR validation of the differentially expressed genes related to regeneration. Thirty nine genes determined as up-regulated
(A) and 8 genes determined as down-regulated (B) in the blastema by the RPKM values were validated using qPCR. Genes are represented
by gene names. Bar values represent mean ± SE of three independent measurements.
study are consistent with those reported previously expression levels of certain genes in blastema might be
[30]. We selected 8 genes classified as down-regulated in attributed to contaminated neighboring tissues. Thus,
blastema to be validated by qPCR. The results confirm in situ hybridization may reveal the specific cells ex-
that 5 of the 8 genes had lower expression levels in blas- pressing the genes in this study. The hybridization
tema. Among them, Wnt-9a and Wnt-2b showed higher showed that patched-2, one of the genes differentially
expression levels in the denervated limb stump than in expressed in blastema, was expressed in the cells lo-
blastema (Figure 5B). cated in blastema (Figure 6A). Hybridization using the
sense control probe showed no signal (Figure 6B). The
Expression patterns of differentially expressed genes in signal of patched-2 expression was limited to only a
limb regeneration portion of blastemal cells, and no signal was shown in
Because of the technical difficulty of completely isolat- the epidermis (Figure 6A). These results confirmed
ing blastema from the underlying stump and epidermis, that patched-2 was differentially expressed in blastema,
we could not exclude the possibility that the higher as seen from NGS and qPCR, and that it was expressedWu et al. BMC Genomics 2013, 14:434 Page 9 of 12
http://www.biomedcentral.com/1471-2164/14/434
Figure 6 In situ hybridization analysis of patched-2 and pax7 expression in the regenerating blastemas. Patched-2 (A) or pax7 (C) signals
(blue) using respective anti-sense probes are shown in the distal tip of regenerating limbs. No signal was detected in the negative-control
sections using respective sense probes of patched-2 (B) and pax7 (D) on serially sectioned slides. The insets in the left upper are the enlarged photos of
respective areas indicated by dotted lines at blastema. Asterisks indicate apical epithelial cap covering the blastema. Scale bars indicate 200 μm.
in blastemal cells rather than contaminated tissues dur- sequences were produced with 39,200 sequences below a
ing tissue sampling. Similarly, pax7 was also shown BLAST cut-off threshold of 10-5. Along with various
expressed in blastemal cells (Figure 6C). Hybridi- existing axolotl transcriptome databases [13-15], these
zation using the sense control probe showed no signal results can further help researchers investigate axolotl’s
(Figure 6D). Hedgehog signaling, the signals mediated limb regeneration. This study also demonstrates the
by activated membrane patched-1 and 2, is required for feasibility of using multiple assemblers to increase the
blastema growth and control of dorsoventral patterning identification accuracy of the differentially expressed genes
[51]. During limb development, hedgehog signaling involved in axolotl limb regeneration.
up-regulates fgf4 to promote limb bud proliferation
[52]. Pax7-positive cells also appear in tail blastemas Methods
and mid bud-stage limb blastemas [42,53]. Pax7 is re- Animal experimental procedures
quired for the specification of myogenic satellite cells, Wild-type and white mutant (d/d) axolotls were kept at
which are myogenic progenitor cells. These results 14–20°C in tap water at pH 6.5-7.5. Juvenile axolotls
demonstrate that NGS procedures can identify poten- with an approximate 80 to 90 mm snout to vent length
tial markers for the blastemal progenitor cells that were subjected to denervation at the brachial plexus
initiate limb regeneration. (third to fifth spinal nerves) of the right upper limbs,
leaving the nerve ends a gap of at least 5 mm. After 1
Conclusion wk, amputation was performed at both forearms at the
This study uses Illumina sequencing technology to se- mid-radius/ulna. At 14 d post-amputation, the full-
quence the axolotl transcriptomes, perform assembly thickness skin at the tip of limb was gently peeled away
and annotation on these sequences, and analyze the dif- by cutting through the full-thickness skin around the
ferentially expressed genes. More than 116,787 distinct circumference of the limb with spring scissors, and theWu et al. BMC Genomics 2013, 14:434 Page 10 of 12
http://www.biomedcentral.com/1471-2164/14/434
blastemas in the left forearms and non-regenerating for functional annotation. The e-value cut-off was set at
stump tissues in the right forearms were collected for 1E-5 for further analysis. Each assembled contig was
total RNA extraction. All the surgical experiments were assigned with the gene name and related function based
conducted under anesthesia with 0.1% MS-222 (Sigma- on the best BLASTx hit (the smallest e-value). Assembled
Aldrich, St. Louis, MO, USA). Animal care and experi- contigs assigned to the same gene were further compared,
mental procedures were approved by the Institutional and the contig from the best e-value was adopted. If there
Animal Care and Use Committee of National Taiwan was a tie between 2 assembled sequences, the one with the
University College of Medicine. largest sequence identity was selected. In the end, with re-
spect to each assembler, a unique contig was used to repre-
RNA extraction and library preparation sent a potential gene of axolotl, and this gene was named
Total RNA was extracted using Trizol Reagent (Invitrogen, by the corresponding protein in Xenopus laevis.
Carlsbad, CA, USA) and isolated by RNeasy mini-columns
(Qiagen, Germantown, MD, USA). RNA quality was Quantization of transcript sequences
assessed by spectrophotometry. A fragmentation buffer Quantitative analysis was performed using CLCbio (CLC
(100 mM ZnCl2 in 100 mM Tris-HCl pH7) was added Genomics Workbench 4.8, parameter settings: minimum
to cut mRNAs into short fragments. Fragmented RNA length fraction, 0.5; minimum similarity fraction, 0.95;
was reverse-transcribed to first-strand cDNA with re- maximum number of hits for a read, 10). The reads
verse transcriptase (Invitrogen) in the presence of a from 2 libraries were mapped to the selected assembled
random hexamer-primer (Invitrogen) and dNTPs for 50 contigs for various assemblers. The read counts accumu-
min at 42°C. The second-strand cDNA was synthesized lated on the selected contigs were further normalized as
using DNA polymerase I in a buffer containing dNTPs RPKM values.
and RNaseH. The short DNA fragments were purified
using the QiaQuick PCR extraction kit (Qiagen) and re- Identification of differentially expressed genes
solved with an elution buffer (10 mM Tris-Cl, pH 8.5) The fold change of RPKM values (the RPKM is replaced
for end reparation and addition of a poly(A) fragment with 0.01 if it equals zero) from the libraries of the two tis-
to both ends. Thereafter, the short fragments were sue types was calculated to identify differentially expressed
connected with sequencing adapters and then separated genes. Because diverse assemblers deliver different sets of
in gels by electrophoresis. The fragments with a size suit- potential axolotl genes, only those homologous to a Xen-
able for NGS were cut from gels and eluted for PCR opus laevis protein and having corresponding contigs from
amplification by using adapter primers. all the 3 assemblers were further analyzed. The genes with
2-fold up- or down-regulation that were consistent among
Analysis of Illumina sequencing results the 3 assemblers were identified.
Four analytic procedures were conducted on the RNA-
seq data. qPCR for mRNA quantification
RNA was prepared using Trizol Reagent (Invitrogen).
Illumina sequencing and de novo assembly The RNA samples of the blastema tissue and dener-
The 2 cDNA libraries were sequenced from both the 5' vated stump end were harvested from the upper limbs
and 3' ends on the Illumina GA II following the manu- 14 days after amputation. The total RNA was reverse-
facturer’s instructions. The low-quality raw sequences transcribed to first-strand cDNA with reverse transcriptase
were removed. The remaining short reads with their (Invitrogen) in the presence of a random hexamer-primer
adaptor sequences trimmed were assembled in a de novo (Invitrogen) and dNTPs for 50 min at 42°C. The expres-
process using 3 assemblers: Trinity [40], trans-ABySS sion levels of specific mRNAs were determined by qPCR
[41], and SOAP de-novo [42]. Similar assembly parame- using respective primer pairs (Additional file 5). Each reac-
ters were used for the 3 assemblers to compare perform- tion was conducted in a total volume of 20 μL containing
ance. Trinity and SOAP de-novo used the default k-mer 50 ng first-strand cDNA, 10 μL 2X Fast SYBR® Green Mas-
setting [40,42], and trans-ABySS was run using k-mer ter Mix (Applied Biosystems, Carlsbad, CA, USA), and
values of 45, 46, to 89 [41]. All assemblies were performed each primer pair at 0.5 μM. A Step One™ Real-Time PCR
on a server with 24 cores and 128 GB of memory. After as- system (Applied Biosystems) was used. Data was analyzed
sembly, the contigs longer than 100 bases were used for using Step One™ software version 2.1 (Applied Biosystems).
subsequent analysis. Error bars indicate RQ max and RQ min.
Functional annotation In situ hybridization
BLASTx [53] was performed to align the assembled Templates of cDNA for axolotl patched-2(319 bp) and
contigs from the 3 assemblers to the nr protein database pax7 (303 bp) were amplified by RT-PCR using respectiveWu et al. BMC Genomics 2013, 14:434 Page 11 of 12
http://www.biomedcentral.com/1471-2164/14/434
sense primers linked to SP6 promoter sequence and anti- genes able to be annotated, which are based for comparison in the
sense primers linked to T7 promoter sequence. Sense and Venn diagrams.
anti-sense RNA probes were synthesized from the cDNA Additional file 5: Primer pairs used in qPCR for validating the
templates using a digoxigenin RNA labeling kit fol- differentially expressed genes.
lowing the manufacturer’s protocol (Roche, Indianapolis,
IN, USA). RNA probes were prepared using respect- Abbreviations
ESTs: Expression sequence tags; NCBI: National Center for Biotechnology
ive primers: patched-2 (anti-sense), 5' CGATTTAGG Information; NGS: Next generation sequencing; qPCR: Quantitative
TGACACTATAGAAGAGTCCAACGACGTGAGGACA polymerase chain reaction; GO: Gene ontology; COG: Clusters of orthologous
AGA- 3'; patched-2 (sense), 5'- CGTAATACGACTC groups; BLAST: Basic local alignment search tool.
ACTATAGGGAGATTGAGCTGGATGGCGTGAA-3'; Competing interests
pax7 (anti-sense), 5'-CGATTTAGGTGACACTATAG We declare that we do not have competing interests.
AAGAGAAACAGGCAGGAGCCAATCAA-3'; and pax7
Authors' contributions
(sense), 5'-CGTAATACGACTCACTATAGGGAGAGCT CHW conducted the molecular genetic studies, participated in the sequence
GACCGGATTCATGTGGTT-3'. alignment, and drafted the manuscript. CCH participated in the sequence
Blastema tissues were fixed overnight at 4°C in 4% alignment. CYC participated in the design of the study and performed the
statistical analysis. MHT conceived the study, participated in its design and
paraformaldehyde in 1× phosphate buffered saline (PBS) coordination, and helped draft the manuscript. HSL designed and
and subsequently embedded in Tissue-Tek (Thermo Sci- coordinated the studies and wrote the manuscript. All authors read and
entific, Miami, FL, USA) and frozen at −80°C until sec- approved the final manuscript.
tioning at 10 μm. Before hybridization, sections were
Acknowledgments
digested with 1 μg/mL proteinase K in 1× PBS at room We thank Mei-Ju Chen, Ting-Ying Chien, Chia-Cheng Hu, Yi-An Tung,
temperature for 8 min, fixed again in 4% paraformalde- Jian Long Huang (Department of Bio-lndustrial Mechatronics Engineering,
hyde in 1x PBS at room temperature for 20 min, and National Taiwan University, Taiwan), and Yu-Chiao Chiu (Graduate Institute of
Biomedical Electronics and Bioinformatics, National Taiwan University, Taiwan) for
carbonated with diethypyrocarbonate in 1× PBS. The their assistance in the sequence alignment. This study was supported by a grant
slides were covered with Hybri-well Press-seal hybridi- (NSC99-2314-B002-107-MY3) from the National Science Council, Taipei, Taiwan.
zation chambers (Sigma-Aldrich) and hybridized over-
Author details
night at 58°C with pre-denatured DIG-labeled probes in 1
Institute of Biotechnology, National Taiwan University, Taipei 106, Taiwan.
2
a hybridization solution (Roche). After hybridization, the Graduate Institute of Biomedical Electronics and Bioinformatics, National
slides were washed in 5× SSC twice, each for 1 h, at 65°C, Taiwan University, Taipei 106, Taiwan. 3Department of Bio-Industrial
Mechatronics Engineering, National Taiwan University, Taipei 106, Taiwan.
and then in 0.1× SSC for 30 min at room temperature. 4
Department of Internal Medicine, National Taiwan University Hospital and
The slides were rinsed in a buffer containing 100 mM National Taiwan University College of Medicine, Taipei 100, Taiwan.
5
Tris-HCl (pH 7.5) and 150 mM NaCl. The slides were Agricultural Biotechnology Research Center, Academia Sinica, Taipei 115,
Taiwan. 6Research Center for Developmental Biology and Regenerative
incubated at 4°C overnight in the same buffer containing Medicine, National Taiwan University, Taipei 100, Taiwan.
an alkaline phosphatase-conjugated anti-digoxigenin
antibody (Roche) diluted at 1:1000. The slides were Received: 30 August 2012 Accepted: 21 June 2013
Published: 1 July 2013
washed in PBST (0.1% TritonX-100 in PBS) 3 times for
30 min each at room temperature. The signals of alka- References
line phosphatase activities bound on the anti-DIG-anti- 1. Brockes JP: Amphibian limb regeneration: rebuilding a complex structure.
Science 1997, 276:81–87.
body were detected using a mixture of the BCIP/NBT 2. Satoh A, Graham GM, Bryant SV, Gardiner DM: Neurotrophic regulation of
solution (Sigma-Aldrich). epidermal dedifferentiation during wound healing and limb
regeneration in the axolotl (Ambystoma mexicanum). Dev Biol 2008,
319:321–335.
Additional files 3. Ferris DR, Satoh A, Mandefro B, Cummings GM, Gardiner DM, Rugg EL: Ex
vivo generation of a functional and regenerative wound epithelium
Additional file 1: Overview of Ambystoma mexicanum transcriptome from axolotl (Ambystoma mexicanum) skin. Dev Growth Differ 2010,
52:715–724.
sequencing and assembly. Length distribution of blastema contigs (A), DN
contigs (B), blastema scaffolds (C), DN transcribed sequences (D), blastema 4. Yang EV, Gardiner DM, Carlson MR, Nugas CA, Bryant SV: Expression of
unigenes (E), DN unigenes (F), and all unigenes (G). DN = denervated Mmp-9 and related matrix metalloproteinase genes during axolotl limb
limb stump. regeneration. Dev Dyn 1999, 216:2–9.
5. Satoh A, Bryant SV, Gardiner DM: Regulation of dermal fibroblast
Additional file 2: Unigenes with a cut-off E-value of 10-5using dedifferentiation and redifferentiation during wound healing and limb
BLASTx against the non-redundant (nr) NCBI nucleotide database. regeneration in the Axolotl. Dev Growth Differ 2008, 50:743–754.
Additional file 3: The 885 unigenes determined up-regulated and 6. Salley-Guydon JD, Tassava RA: Timing the commitment to a
1068 unigenes determined down-regulated in blastema. wound-healing response of denervated limb stumps in the adult newt,
Additional file 4: Comparisons of the identified differentially Notophthalmus viridescens. Wound Repair Regen 2006, 14:479–483.
expressed genes between this study and previous sequencing and 7. Bryant SV, Endo T, Gardiner DM: Vertebrate limb regeneration and the
microarray data. The “Up” and “Down“ spreadsheets include all 885 origin of limb stem cells. Int J Dev Biol 2002, 46:887–896.
8. Gardiner DM, Endo T, Bryant SV: The molecular basis of amphibian limb
up-regulated and 1068 down-regulated genes, respectively. The “Up
signature” and “Down signature” spreadsheets include only those regeneration: integrating the old with the new. Semin Cell Dev Biol 2002,
13:345–352.Wu et al. BMC Genomics 2013, 14:434 Page 12 of 12
http://www.biomedcentral.com/1471-2164/14/434
9. Vera JC, Wheat CW, Fescemyer HW, Frilander MJ, Crawford DL, Hanski I, 34. Koshiba K, Kuroiwa A, Yamamoto H, Tamura K, Ide H: Expression of Msx
Marden JH: Rapid transcriptome characterization for a nonmodel genes in regenerating and developing limbs of axolotl. J Exp Zool 1998,
organism using 454 pyrosequencing. Mol Ecol 2008, 17:1636–1647. 282:703–714.
10. Parchman TL, Geist KS, Grahnen JA, Benkman CW, Buerkle CA: 35. Torok MA, Gardiner DM, Izpisua-Belmonte JC, Bryant SV: Sonic hedgehog
Transcriptome sequencing in an ecologically important tree species: (shh) expression in developing and regenerating axolotl limbs. J Exp Zool
assembly, annotation, and marker discovery. BMC Genomics 2010, 11:180. 1999, 284:197–206.
11. Animal Genome Size Database [http://www.genomesize.com] 36. Carlson MR, Komine Y, Bryant SV, Gardiner DM: Expression of Hoxb13 and
12. Yamada K, Lim J, Dale JM, Chen H, Shinn P, Palm CJ, Southwick AM, Wu HC, Hoxc10 in developing and regenerating Axolotl limbs and tails. Dev Biol
Kim C, Nguyen M, et al: Empirical analysis of transcriptional activity in the 2001, 229:396–406.
Arabidopsis genome. Science 2003, 302:842–846. 37. Tamura K, Yonei-Tamura S, Yano T, Yokoyama H, Ide H: The autopod: its
13. Habermann B, Bebin AG, Herklotz S, Volkmer M, Eckelt K, Pehlke K, Epperlein HH, formation during limb development. Dev Growth Differ 2008, 50:S177-–187.
Schackert HK, Wiebe G, Tanaka EM: An Ambystoma mexicanum EST Suppl 1.
sequencing project: analysis of 17,352 expressed sequence tags from 38. Zakany J, Duboule D: The role of Hox genes during vertebrate limb
embryonic and regenerating blastema cDNA libraries. Genome Biol 2004, development. CurrOpin Genet Dev 2007, 17:359–366.
5:R67. 39. Koniski A, Cohen N: Axolotl (Ambystoma mexicanum) lymphocytes
14. Monaghan JR, Epp LG, Putta S, Page RB, Walker JA, Beachy CK, Zhu W, Pao GM, produce and are growth-inhibited by transforming growth factor-beta.
Verma IM, Hunter T, et al: Microarray and cDNA sequence analysis of Dev Comp Immunol 1998, 22:91–102.
transcription during nerve-dependent limb regeneration. BMC Biol 2009, 7:1. 40. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, Adiconis X,
15. Huggins P, Johnson CK, Schoergendorfer A, Putta S, Bathke AC, Stromberg Fan L, Raychowdhury R, Zeng Q, et al: Full-length transcriptome assembly
AJ, Voss SR: Identification of differentially expressed thyroid hormone from RNA-Seq data without a reference genome. Nat Biotechnol 2011,
responsive genes from the brain of the Mexican Axolotl (Ambystoma 29:644–652.
mexicanum). Comp Biochem Physiol C Toxicol Pharmacol 2012, 155:128–135. 41. Robertson G, Schein J, Chiu R, Corbett R, Field M, Jackman SD, Mungall K,
16. SGP/Sal-Site [http://www.ambystoma.org] Lee S, Okada HM, Qian JQ, et al: De novo assembly and analysis of
17. Monaghan JR, Athippozhy A, Seifert AW, Putta S, Stromberg AJ, Maden M, RNA-seq data. Nat Methods 2010, 7:909–912.
Gardiner DM, Voss SR: Gene expression patterns specific to the 42. Li R, Yu C, Li Y, Lam TW, Yiu SM, Kristiansen K, Wang J: SOAP2: an
regenerating limb of the Mexican axolotl. Biol Open 2012, 1:937–948. improved ultrafast tool for short read alignment. Bioinformatics 2009,
18. Campbell LJ, Suarez-Castillo EC, Ortiz-Zuazaga H, Knapp D, Tanaka EM, 25:1966–1967.
Crews CM: Gene expression profile of the regeneration epithelium 43. Kumar A, Godwin JW, Gates PB, Garza-Garcia AA, Brockes JP: Molecular
during axolotl limb regeneration. Dev Dyn 2011, 240:1826–1840. basis for the nerve dependence of limb regeneration in an adult
19. Holt RA, Jones SJ: The new paradigm of flow cell sequencing. Genome Res vertebrate. Science 2007, 318:772–777.
2008, 18:839–846. 44. Miyazaki K, Uchiyama K, Imokawa Y, Yoshizato K: Cloning and characterization
20. Shendure J, Ji H: Next-generation DNA sequencing. Nat Biotechnol 2008, of cDNAs for matrix metalloproteinases of regenerating newt limbs. Proc Natl
26:1135–1145. Acad Sci U S A 1996, 93:6819–6824.
21. Mardis ER: Next-generation DNA sequencing methods. Annu Rev Genomics 45. Zeng L, Kempf H, Murtaugh LC, Sato ME, Lassar AB: Shh establishes an
Hum Genet 2008, 9:387–402. Nkx3.2/Sox9 autoregulatory loop that is maintained by BMP signals to
22. Li R, Zhu H, Ruan J, Qian W, Fang X, Shi Z, Li Y, Li S, Shan G, Kristiansen K, induce somatic chondrogenesis. Genes Dev 2002, 16:1990–2005.
et al: De novo assembly of human genomes with massively parallel 46. Schnapp E, Kragl M, Rubin L, Tanaka EM: Hedgehog signaling controls
short read sequencing. Genome Res 2010, 20:265–272. dorsoventral patterning, blastema cell proliferation and cartilage
23. Pertea G, Huang X, Liang F, Antonescu V, Sultana R, Karamycheva S, Lee Y, induction during axolotl tail regeneration. Development 2005,
White J, Cheung F, Parvizi B, et al: TIGR Gene Indices clustering tools 132:3243–3253.
(TGICL): a software system for fast clustering of large EST data sets. 47. Kumar A, Velloso CP, Imokawa Y, Brockes JP: The regenerative plasticity of
Bioinformatics 2003, 19:651–652. isolated urodele myofibers and its dependence on MSX1. PLoS Biol 2004,
24. Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M: The KEGG resource 2:E218.
for deciphering the genome. Nucleic Acids Res 2004, 32:D277–D280. 48. Robert B, Sassoon D, Jacq B, Gehring W, Buckingham M: Hox-7, a mouse
homeobox gene with a novel pattern of expression during
25. Dungan KM, Wei TY, Nace JD, Poulin ML, Chiu IM, Lang JC, Tassava RA:
embryogenesis. EMBO J 1989, 8:91–100.
Expression and biological effect of urodele fibroblast growth factor 1:
49. Song K, Wang Y, Sassoon D: Expression of Hox-7.1 in myoblasts inhibits
relationship to limb regeneration. J Exp Zool 2002, 292:540–554.
terminal differentiation and induces cell transformation. Nature 1992,
26. Poulin ML, Patrie KM, Botelho MJ, Tassava RA, Chiu IM: Heterogeneity in
360:477–481.
the expression of fibroblast growth factor receptors during limb
50. Muragaki Y, Mundlos S, Upton J, Olsen BR: Altered growth and branching
regeneration in newts (Notophthalmus viridescens). Development 1993,
patterns in synpolydactyly caused by mutations in HOXD13.
119:353–361.
Science 1996, 272:548–551.
27. Poulin ML, Chiu IM: Nucleotide sequences of two newt (Notophthalmus
51. McHedlishvili L, Epperlein HH, Telzerow A, Tanaka EM: A clonal analysis of
viridescens) fibroblast growth factor receptor-2 variants. Biochim Biophys
neural progenitors during axolotl spinal cord regeneration reveals
Acta 1994, 1220:209–211.
evidence for both spatially restricted and multipotent progenitors.
28. Poulin ML, Chiu IM: Re-programming of expression of the KGFR and bek
Development 2007, 134:2083–2093.
variants of fibroblast growth factor receptor 2 during limb regeneration
52. Laufer E, Nelson CE, Johnson RL, Morgan BA, Tabin C: Sonic hedgehog and
in newts (Notophthalmus viridescens). Dev Dyn 1995, 202:378–387.
Fgf-4 act through a signaling cascade and feedback loop to integrate
29. Giampaoli S, Bucci S, Ragghianti M, Mancino G, Zhang F, Ferretti P:
growth and patterning of the developing limb bud. Cell 1994, 79:993–1003.
Expression of FGF2 in the limb blastema of two Salamandridae
53. Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ:
correlates with their regenerative capability. Proc Biol Sci 2003,
Gapped BLAST and PSI-BLAST: a new generation of protein database
270:2197–2205.
search programs. Nucleic Acids Res 1997, 25:3389–3402.
30. Satoh A, Makanae A, Hirata A, Satou Y: Blastema induction in aneurogenic
state and Prrx-1 regulation by MMPs and FGFs in Ambystoma
doi:10.1186/1471-2164-14-434
mexicanum limb regeneration. Dev Biol 2011, 355:263–274.
Cite this article as: Wu et al.: De novo transcriptome sequencing of
31. Gardiner DM, Blumberg B, Bryant SV: Expression of homeobox genes in
axolotl blastema for identification of differentially expressed genes
limb regeneration. Prog Clin Biol Res 1993, 383A:31–40. during limb regeneration. BMC Genomics 2013 14:434.
32. Gardiner DM, Blumberg B, Komine Y, Bryant SV: Regulation of HoxA
expression in developing and regenerating axolotl limbs.
Development 1995, 121:1731–1741.
33. Torok MA, Gardiner DM, Shubin NH, Bryant SV: Expression of HoxD genes
in developing and regenerating axolotl limbs. Dev Biol 1998, 200:225–233.You can also read
